f t i +

Conversational Marketing & Bot Theory: A Video Interview with HubSpot's Brian Bagdasarian

Mike Skeehan
Last month, I had the privilege of meeting Brian Bagdasarian, Head of Conversational Growth Strategy at HubSpot, about chatbots and conversational marketing. Before he presented at our #HUGPasadena event, he sat down for an interview to discuss the theory behind bot narratives, best practices and use cases, and the state of product.

We were sure to discuss his background and adult beverage preferences as well. Enjoy the interview, and be sure to check out our Intro to Chatbots primer.



 Mike: Thanks for coming! Why don't you start. Why don't we start by just, give an overview of who you are, kind of what your role is, what your recent history has been.

Brian: Yeah, sure. So, my name is Brian Bagdasarian. There's 11 letters in that last name. I think there are five vowels, fun fact. I joined HubSpot about a year and a half ago through the acquisition of Motion AI, which was a chatbot development platform.

At the time we were actually bought, we were the largest out there. We had 34,000 users and 73,000 bots built on the platform. To put some perspective on that, that is bigger than anybody, even today that is still out there. ManyChat is getting right around there, but we were huge, and we were a four-person team, so that's insane. We didn't sleep.

But, HubSpot came to us and it was a good fit for the two companies. And I joined on. And David, who was the founder of the company came on. And he now runs Bots, and I went more to the educational and theory side to carry over some of the work that I've been doing.

And I've been the head of conversational growth strategy for, you know, for the last year and a half or so. Really, kind of just trying to teach people how to have conversations and how to think about them in a logical and programmatic way, but that's still very accessible.

Mike: Okay. Well, let's start there. It seems like your passion is really, figuring out, breaking down language conceptually, and programmatically as it relates to application in messaging. Is that accurate?

Brian: Yeah, I mean, look, if you think about something like ordering a pizza, all right?

When you order a pizza, there's a whole bunch of parts to that process. It's pretty complex when you think about it. You've got to figure out where they are, who they are, what they want, what they need, how long's it take, all that stuff.

But, if you think about it from a very simple standpoint, that macro conversation is made up of these micro conversations. And micro conversation's are really powerful, because they have a beginning, middle, and an end. When they have a beginning, middle, and end, you can then measure how many were started? How many reached the middle point? And how many reached the end point?

If you can measure it, you can determine success from it. So, think about this: What's the first thing that, when you call the pizza place and you say, "I need to order a pizza," what do they say to you?

Mike: They'll ask you if it's for carryout or delivery.

Brian: Why do they ask you that?

Mike: Well, the answer to that question dictates the next sequence of questions.

Brian: Right, right. So, you say, "Hi, I'd like to order a pizza." Or, "I'd like to place an order." They say, "Is it for delivery or carry out?" You say, "carry out," they don't have to ask you where you are. So, the end of that conversation, that micro conversation is the determiner of is it for delivery or carryout.

If you said it was for delivery, then they have to go to a different part of the conversation, which is, "Okay, can I have your address?" Or your zip code, or whatever it may be. But, if it's for carryout, you don't have to have that conversation, so you move down to the next step, which is taking the order.

These are all micro conversations. If I said, "Hello," and you said ...

Mike: Hey, how's it going?

Brian: It's going great. My name's Brian.

Mike: Yeah, I'm Mike.

Brian: Mike ... These are conditioned responses. That's what we want to mimic, are these conditioned responses. Because they sound natural. But, once I have your name and you have my name, I can reuse that information later on. And that makes it sound personal.

And if I'm coming at you in a cheerful mood, and you come back in a cheerful mood, we're staying in a relatively cheerful mood. We are matching the sentiment for each other. But, as we go through the process and we start to ask questions. And you ask me a question and I give you answers, there's really only eight ways you can ask a question...

Mike: Like, literally? Okay.

Brian: Literally eight ways.

Mike: Yeah, okay.

Brian: Who, what, where, when, why, how, which, and yes/no, which is can I, will I, should I, or do I, for instance. That's it. There's literally no other way to ask a question. And every one of those, with some variations, I mean you can get more and more complex if you want to, has a kind of answer that you're looking for.

And it's the idea of treating bot building not as a technical process, but as a virtual HR process.

Mike: Elaborate on that.

Brian: Okay, well, you've hired someone before, right?

Mike: Yep.

Brian: So, before you go and hire somebody, and we're talking relatively low level, right? Low level person. Before that person comes in, what do you have to put together for that person's first day?

Mike: An orientation packet.

Brian: Yeah, information about the company. Ways to talk about the company, what the company does, where it's located, how long it's been around. All those things, right? Standard answers, essentially -- standard information.

Okay, now when that person who, let's say they're gonna be manning your live chat. This is a real person we're talking about here. The next thing you'd probably do is teach them how to answer common questions.

Take the standard information that they have available, and then, here's how to answer these questions. Here's the steps to answer these different types of questions, but the information fundamentally is standardized.

It doesn't mean that it's not always the answer. It just means that those answers are not changing because the wind's blowing differently.

Mike: Sure, there's basic templates around which the answers are built, yep.

Brian: Right, then you get into optimizing. You probably teach them how to best and most effectively answer that. If it's on a live chat, don't send them over four pages of typed information.

In fact, it's one of the ways that's really easy to grab someone's email address. You know, a lot of chat bots out there, a lot of people in live chat, the first thing they do is like, "Hi, what's your email?"

No. I don't want to give you my email. I don't have any trust in you. I don't know you.

Mike: Yeah, now you just feel like a form masquerading as a conversation.

Brian: Right, right, and that's just no, and we'll come back to that. Instead, if I, if you said, you know, "Hey, I need some information on the company. Can you tell me what you guys do?"

“Sure, I can actually send you over some information on this. What's your email address?”

That's a pretty natural way to get your email address, but I now just have gotten it from you, and you're gonna give it to me, because...

Mike: You want the information and...

Brian: And it's a logical way for me to deliver it to you, because it's too long to give you right then. Okay, now if it was a one or two sentence answer, I can just shoot it to you right back over in the live chat. Or, if I already had your email address, I might just send it to you in a link. And I can actually continue that conversation on that new page, depending on where you are, right? Make sense?

Mike: Right, yep, it does.

Brian: So, now you have this new hire. You've taught them the basic information on the company. You've taught them how to answer within the context of a given point of a conversation. You've instructed them on when to switch what channel they're delivering information on. You know, from live chat to email, or whatever it may be. Would you expect that person to remember everything about every person they ever talked to, forever?

Mike: Oh yeah, certainly not.

Brian: No. So, what would you expect them to be able to do?

Mike: Well, document the information in a way that's easily accessible.

Brian: Like, in a ...

Mike: CRM.

Brian: Right. You teach somebody to use a CRM. So, when they get more information, they can do that. Also, to access the information that is already had.

Right? And then, you probably want to teach them how to be empathetic. How to be human. How to deal with somebody that might be upset, or match somebody if they're very happy.

Mike: Sure.

Brian:  Well, that's the basis of SCOPE. Those are the principles of conversational growth strategy. Those are also the principals of inbound.

Mike: Now, SCOPE is an acronym?

Brian: It's an acronym. Standardize, Conceptualize, Optimize, Personalize, and Empathize. It's the exact same way that you would take somebody normal, a human, through that training process.

Mike: You're gonna do the same thing with a bot.

Brian: Right.

Mike: What percentage of conversations a standard organization has, or questions that a standard organization would field, do you think can be translated into… [a narrative]

Brian: I mean, I would say that you can probably get 90% of the most common ones.

Mike: Yep.

Brian: Conversational growth is not just about building chatbots. It's about building a better way to develop a one to one relationship with somebody that is based on trust, that is based on empathy, that is based on basically, humanity.


"Conversational growth is not just about building chatbots. It's about building a better way to develop a one to one relationship with somebody that is based on trust, that is based on empathy, that is based on basically, humanity."

It just doesn't mean that the other side of the conversation is always gonna be an actual human.

Mike: There's a spectrum of chat-related services out there, or chat-related apps out there. The concepts that you've just articulated, how well are the current products, options that are on the market, integrating those concepts, enabling organizations to succeed, and how do you see that evolving in the short to medium term?

Brian: So, one of the things I didn't mention at all is artificial intelligence.

Mike: Yep.

Brian: Because you don't need it. For the vast majority of conversations, you don't need AI. You just need to understand the conversational narrative, what you're trying to do.

With that being said, you know, HubSpot Conversations, a few other ones, ManyChat can do this. Drift can do it. You can take the information you're collecting, and you can drop it into your CRM. And that's a really big part of it. The rest of it has nothing to do with actually, what you're using for a platform. It has to do with how you're using the platform.

Mike: What will serve in your estimation to differentiate these different applications, these options on the market, as time progresses? If AI is not really required, how are they gonna differentiate from each other?

Brian: The ability to integrate.

Mike: Okay.

Brian: The ability to create this idea of what we call shared knowledge pool. So, have you ever called up a place, you know, say, you call two times...

Mike: And they don't remember the second call, the first call.

Brian: Anything about you. It would be the equivalent, it's like conversational face blindness. Imagine if in the morning, you left your house. You kissed your wife goodbye and said, "Hey, I'm going to work." She goes, "Great, here you go." That was it, you recognize her. When you come home, she has no clue who you are.

Mike: Yep.

Brian: That'd be frustrating, right? After a week you'd be like, "What the heck is going on here?" Yet, that seems to be acceptable for the vast majority of businesses, right?

So, how do you differentiate? I should never have to ask you a question twice. And the platform that lets me do that is the one that's gonna win.

Mike: Yep, so the CRM underpinnings are super important.

Brian: It's not just CRM. It's also about having access to answers. It's not just about what they know about me. It's about how easily I can access the information that I need. Statistically speaking, 72-ish percent of people do not care if they are getting their answer from a chat bot or from a live person, as long as they get their answer. In a reasonable period of time. Live chat, that's anywhere from 30 seconds to five minutes.

Mike: Okay.

Brian: After five minutes for the first time ... So, if I send you a message, if I wait more than five minutes, I'm starting to get ticked off. If we're in the middle of the conversation, I'm waiting more than 30 seconds or a minute, I'm starting to get ticked off. So “time to live” is that concept.

Mike: Okay. Two more questions: One, if you were a marketer, being introduced, beginning to explore conversational marketing, where do you see the most opportunity, looking forward for marketers?

Brian: Specific solutions.

Mike: Okay.

Brian: It's not building the next big platform. It's figuring out what ... Okay, let me break this down. Every business has business processes. So, the first thing as a marketer, you got to be able to figure out: what is that business - that client business, what's their process look like?

Then, you got to figure out where the bottleneck exists. Every business process has a point where it becomes dramatically inefficient. And that's the bottleneck. Figure out what conversation is had there. Map that conversation out. Can that be automated? What are the measurement points of that?

Mike: Sure, yep.

Brian: How can I simplify that? If you go to a business and say, "Listen, I can save you 50 man hours a month because we're gonna automate just a specific point in this one business process, it's a pretty big value prop.

If I was a marketer, I would be concentrating on becoming an expert at understanding the client businesses, business process, and then be aiming to coach them through the process of sort of, reverse engineering, what that looks like.

Mike: What can be automated.

Brian: And once you do that, you can then build forward.

Mike: Cool. Last question. You're from Maine.

Brian: Yeah.

Mike: What is your favorite spirit to keep you warm on a cold Maine snowy night?

Brian: That'll keep me warm? That's a good one.

Mike: A fire in the fireplace, snow falling outside, and you're sipping on a...

Brian: On a liter of Fireball! No, it's not that. No, actually, there's a company called Maine Craft Distilling. And they do this, it's called a botanical spirit. It's mostly carrots.

Mike: All right.

Brian: And it has the same proof as, like, a vodka. It is unbelievable.

Mike: Okay, you like how I just projected my own preconceived notions about what it's like to live in Maine? You're in a cabin with a fireplace, sipping on a warm drink on a snowy night...

Brian: In the winter, you very likely are! In the summer, you're on a boat doing it.

Mike: Oh, good, I got it. Nailed it, yeah.

Brian: But you know what I'm actually a huge fan of? Hard seltzer. I am telling you, there is nothing better than being able to drink a fizzy seltzer water that'll also get you lit.

Mike: Yeah, you get hydrated, and you...

Brian:  Oh my God, I mean, it's fantastic. You know, you give me a sixer of black cherry hard seltzers, I am a happy, happy, very confident man.

Mike: Perfect, yeah. Well, thanks for swinging in.

Brian: Of course, man.

Both: Cheers!

Need help refining your chatbot? Schedule a consultation with our team and receive a free assessment when you mention this interview. 


Learn more about Conversational Marketing best practices by subscribing to our blog newsletter (below) and contact our team to learn more about bot building, tools for success, and "Showing Up" for your users or clients.
Definitely not spam

Sign up for our newsletter

Don't worry - we only average, like, two emojis per subject line.

Got a question for Mike Skeehan?

Message the author of this post and they'll get back to you.

Fire Away